Cursor Launches a New AI Agent Experience to Take On Claude Code and Codex

Cursor Debuts Cursor 3: The Agent-First Coding Interface

Cursor, a well-known AI coding startup, has announced Cursor 3, their latest generation product aimed at challenging dominant offerings from OpenAI and Anthropic. The upgraded platform allows users to deploy dedicated AI coding agents to handle tasks, focusing on agentic workflows that have become popular among developers.

A New Way to Code: From IDEs to Agents

The Cursor 3 interface takes a major leap by enabling developers to request coding features or fixes through simple natural language instructions in a chatbot-like environment. Agents then carry out the work, letting engineers review and manage all active agents within the same desktop app. This seamless integration differentiates Cursor 3 from competitors, by blending agent-driven processes with Cursor’s established developer environment.

Competing With Deep-Pocketed AI Labs

As AI-powered agent tools like Claude Code and Codex continue to grow in popularity—helped by aggressive pricing—Cursor faces pressure to keep up. Recent changes to subscription models and the shift toward usage-based pricing were deemed necessary for Cursor’s sustainability. In response, the startup is investing in its own AI models, including the recently introduced Composer 2, to deliver cost-effective, capable solutions.

Looking Ahead

Cursor’s transition to agent-first coding represents a bet on the future of software development, where AI agents handle much of the heavy lifting for programmers. With significant investments and competition, Cursor’s ability to evolve will be measured by how quickly it can raise capital and expand its technology.

For further details, read the original article by Maxwell Zeff on WIRED.